Fig. 3

Associations of variability in sleep duration (A), sleep midpoint (B), and sleep offset (C) (calculated as standard deviation, SD, of hours) per adolescent with average probability of breakfast consumption (between-person associations) in three separate linear mixed models. Each predictor is represented by each adolescent’s SD across all time points. The mean number of valid actigraphy nights was 5.6 ± 1.4 (range: 3–9; IQR 5–7) and the mean number of breakfast reports was 5.5 ± 1.4 (range: 3–9; IQR 4–7) per adolescent. All models adjust for sleep duration (linear and quadratic, sleep duration²) and demographic/household covariates: school day, boredom, loneliness, happiness, birth sex, race/ethnicity, household income, body mass index percentile, and depressive symptoms. Shaded bands depict 95% confidence interval of probability of breakfast consumption predicted from each sleep measure